Systematic Review of the Literature on Pain in Patients with Polytrauma Including Traumatic Brain Injury

Abstract

Objective

To review the literature addressing the assessment and management of pain in patients with polytraumatic injuries including traumatic brain injury (TBI) and blast-related headache, and to identify patient, clinician and systems factors associated with pain-related outcomes.

Design

Systematic review.

Methods

We conducted searches in MEDLINE of literature published from 1950 through July 2008. Due to a limited number of studies using controls or comparators, we included observational and rigorous qualitative studies. We systematically rated the quality of systematic reviews, cohort, and case-control design studies.

Results

One systematic review, 93 observational studies, and one qualitative research study met inclusion criteria. The literature search yielded no published studies that assessed measures of pain intensity or pain-related functional interference among patients with cognitive deficits due to TBI, that compared patients with blast-related headache with patients with other types of headache, or that assessed treatments for blast-related headache pain. Studies on the association between TBI severity and pain reported mixed findings. There was limited evidence that the following factors are associated with pain among TBI patients: severity, location, and multiplicity of injuries; insomnia; fatigue; depression; and post-traumatic stress disorder.

Conclusions

Very little evidence is currently available to guide pain assessment and treatment approaches in patients with polytrauma. Further research employing systematic observational as well as controlled intervention designs is clearly indicated.

Introduction

Major advances in body armor technology and battlefield medicine have improved survival from combat injuries that would have been fatal in previous wars [1]. Data from the Department of Defense (DOD) indicate that the lethality of injuries has decreased from 24% among those wounded in the Vietnam War, to 10% among those wounded in the Operation Enduring Freedom/Operation Iraqi Freedom (OEF/OIF) conflicts [2]. Soldiers wounded in our nation's current wars have more multiple and complex injuries and emotional trauma than typically seen in past wars [3,4]. Blast exposure was responsible for 71.5% of traumatic brain injuries and 49.9% of amputations among service members as of August 2008 [5]. Auditory or visual impairments, spinal cord injury (SCI), post-traumatic stress disorder (PTSD), and other mental health conditions are also common.

In 2005, polytrauma was defined by the Veterans Health Administration (VHA) as “injury to the brain in addition to other body parts or systems resulting in physical, cognitive, psychological, or psychosocial impairments and functional disability” [6]. VHA's definition of polytrauma has since expanded to encompass concurrent injury to two or more body parts or systems resulting in cognitive, physical, psychological, or other psychosocial impairments.

Pain resulting from polytraumatic injuries poses numerous challenges during and following rehabilitation. It is unclear how best to assess and monitor pain among patients with cognitive deficits due to traumatic brain injury (TBI). Due to cognitive side effects, commonly used pain treatments (for example, oral opioids) have the potential to interfere with the active rehabilitation needed to restore function in this population. Blast-related headache may differ in terms of phenomenology and treatment from other types of headache.

The objectives of this study were to systematically review the empirical literature to address the assessment and management of pain in patients with polytraumatic injuries, to identify patient, clinician and systems factors associated with pain-related outcomes in polytrauma patients, and to outline a research agenda to address key questions.

Methods

In consultation with a technical advisory panel (clinical or research experts working in the areas of pain, polytrauma or TBI, primarily within the Veterans Affairs (VA) system), we identified five key questions for the review. These questions addressed 1) methods of assessment of pain among patients with cognitive deficits due to TBI; 2) effectiveness of treatment approaches for pain related to polytrauma; 3) the phenomenology and management of blast-related headache; 4) patient factors associated with pain-related outcomes among polytrauma patients; and 5) clinician and systems factors associated with pain-related outcomes among polytrauma patients. The specific key questions are included with the results below.

Polytrauma was defined for this review as concurrent injury to two or more body parts or systems resulting in cognitive, physical, psychological, or other psychosocial impairments. Consistent with VHA's definition, TBI of moderate or greater severity was considered polytrauma (head injury itself plus associated cognitive sequelae). Combat-related mental conditions co-occurring with injury to at least one other system also constituted polytrauma.

The scope of this review included the assessment and treatment in rehabilitation and post-rehabilitation care settings of persistent pain or exacerbations of pain resulting from polytraumatic injuries. We included studies measuring pain-related outcomes, specifically pain intensity and pain-related function or interference, 3 months or more from the date of injury. Studies examining battlefield/emergency or assessment and care within 3 months of injury were not included unless they also examined pain outcomes 3 months following injury. We also did not include studies examining choice of specific surgical strategy, perioperative management of traumatic (including burn) injuries, or use of particular procedures or devices for specific orthopedic injuries. We excluded studies describing functional outcomes of polytrauma unless a pain measure was also included. Finally, we excluded studies of post-traumatic/post-concussive headache unless the sample included patients with moderate or severe head injury or included a majority of patients with blast-related head injury. There have been a number of narrative reviews of assessment and treatment of post-traumatic headache among patients with mild-TBI or post-concussive syndrome; we felt that inclusion of these studies was beyond the scope of our key questions. Figure 1 illustrates the analytic framework that guided our review.

Figure 1. Analytic framework.

Two research librarians independently designed search strategies based on the key questions, and conducted searches in PubMed and Ovid MEDLINE of literature published from 1950 through July 2008. Appendix A provides the search strategies in detail. The results of both searches were combined into a single reference library. Additional articles were identified from reference lists of studies, review articles, editorials, and by consulting experts; some of these articles were published prior to 1950. We also searched for relevant studies in the following databases: PsychINFO; the PILOTS Database (the VA PTSD bibliographic database); REHABDATA, the bibliographic database of the National Rehabilitation Information Center; DOD Technical Information Center; and the Cochrane Database of controlled clinical trials. All citations were imported into an electronic database (EndNote X1).

Three investigators (SD, RC, MF) reviewed the titles and abstracts identified from the searches. Full-text articles of potentially relevant abstracts were retrieved for further review. Reference lists from pertinent articles were reviewed to find additional articles for inclusion. Eligible articles had English-language abstracts and provided primary data relevant to the key questions. For a study to be eligible for Key Questions 1, 2, 4, and 5, the sample had to have all or a majority of patients with polytrauma, or analyses and findings had to be stratified by whether the patients had polytrauma, such that readers could discern outcomes for the polytrauma group.

Eligible study designs included controlled clinical trials, systematic reviews, as well as prospective and retrospective cohort studies, case-control design studies, and qualitative studies using rigorous qualitative research methods. Due to a limited number of studies that included a comparator group, we also considered relevant cross-sectional and case report/case series studies for inclusion for some of the key questions. To rate the quality of studies we used criteria developed by the U.S. Preventive Services Task Force for rating randomized controlled trials, cohort studies, and case control studies (Appendix B) [7]. We did not rate the quality of cross-sectional studies, case reports, or case series. We assessed the overall quality of evidence for outcomes using a method developed by the Grading of Recommendations, Assessment, Development, and Evaluation (GRADE) Working Group (Appendix C) [8], which classified the grade of evidence across outcomes according to the following criteria: High = Further research is very unlikely to change our confidence on the estimate of effect; Moderate = Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate; Low = Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate; Very Low = Any estimate of effect is very uncertain. Finally, a draft version of our findings was sent to technical reviewers who provided comments, suggested additional pertinent references, and prioritized future research topics and study designs.

Results

The combined library contained 3,252 citations, of which we reviewed 578 articles at the full-text level. From these, we identified systematic reviews and observational studies that addressed one or more of the key questions. Figure 2 shows the results of the literature search and the organization of themes that emerged for each key question.

Figure 2. Management of pain in polytrauma literature flow.

Key Question 1

Have reliable and valid measures and assessment tools been developed to measure pain intensity and pain-related functional interference among patients with cognitive deficits due to TBI? Which measures and tools are likely to be most useful in assessing pain in polytrauma patients with cognitive deficits due to TBI?

Findings

We found no published studies that assessed reliability and validity of measures of pain intensity or pain-related function among patients with cognitive deficits due to TBI.

Key Question 2

Which treatment approaches are most likely to be effective in improving pain outcomes (pain intensity and functional interference) in polytrauma patients? Which pain treatment approaches are most likely to enhance overall rehabilitation efforts?

Findings

We found no randomized controlled trials, systematic reviews, prospective cohort, case-control, or systematic observational studies that tested the efficacy or effectiveness of specific pain treatment approaches among patients with polytrauma. One study examined the effectiveness of inpatient rehabilitation in helping patients with trauma-related amputation return to work [9]. This study was a fair-quality retrospective cohort study of patients from an urban trauma center between 1984 and 1994. Seventy-eight patients who had undergone trauma-related amputation were contacted and interviewed approximately 7 years after injury. The study excluded amputation for non-injury reasons (e.g., diabetes), SCI or TBI. Many patients had multiple injuries. After controlling for demographic factors, injury characteristics, and other medical morbidity, inpatient rehabilitation was marginally associated with increased likelihood of return to work (P = 0.09) and decreased likelihood of reduced hours of work (P = 0.05) (GRADE: Very Low).

We also identified a number of case reports and case series that described pain treatment approaches and pain outcomes among patients with polytrauma including TBI [10–31]. Several case reports supported that intrathecal baclofen may be helpful for spasticity associated with TBI and related injuries [21–25]. In patients with complex regional pain syndrome I, topical capsaicin brought pain relief in a single-case study [13], and early intervention with spinal cord stimulation was associated with decreases in pain and opioid use in a consecutive-case series of 10 active duty U.S. military personnel, 6 who were injured in OEF/OIF [30].

Key Question 3

Does blast-related headache pain differ in terms of phenomenology and treatment from other types of headache pain? Which treatments are best for persistent blast-related headache pain?

Findings

We found no randomized controlled trials, cohort studies, case-control studies, or other systematic observational studies that compared patients with blast-related headache with patients with other types of headache or that specifically addressed treatments for blast-related headache pain.

Key Question 4

What patient factors are associated with better and worse (pain-related) clinical outcomes among polytrauma patients? Have interventions been developed to specifically address these factors?

Findings—Patients with TBI

We found no randomized controlled trials. One systematic review, 11 cohort, 2 case-control, and 17 cross-sectional studies specifically addressed patient factors associated with outcomes in TBI patients. Table 1 provides a summary of the data abstracted from the systematic review and cohort studies. One fair-quality systematic review examining relationships between TBI severity and pain was published in 2008 [32]. The review identified 23 studies (15 cross-sectional, 5 prospective observational, and 3 retrospective observational) including 4,206 patients. No randomized clinical trials were identified. Data were pooled across studies utilizing varying samples and designs to obtain overall prevalence rates. The review showed that 58% of patients with TBI have chronic headache. Brain injury was associated with headache even after adjustment for PTSD [32].

Table 1. Key Question 4 Studies: Patients with traumatic brain injury (TBI).

Study	Design and Sample	Pain-Related Outcomes	Analytic Method	Results
Brenner 1944 [55]	Prospective cohort; 200 inpatients with head injury	Headache prevalence every 2 months up to 15 months	Descriptive	Sixty-two percent of patients reported headaches in hospital and 41% reported headaches after hospital. The prevalence of headache persisting longer than 2 months was similar across severity of head injury
Bryant et al. 1999 [35]	Prospective cohort; 96 TBI patients admitted to inpatient rehab. facilities (Australia)	10 pt VAS-pain intensity, location, frequency; 6 months	Bivariate comparisons	More patients who reported chronic pain (37%) met criteria for PTSD than did those without pain (15%). Frequency but not severity of chronic pain associated with PTSD
Bushnik et al. 2008 [60]	Prospective cohort; 38 TBI patients admitted to inpatient rehab. facility	VAS 10 pt. scale; 1 &2 years; measure not a primary study outcome	Bivariate comparisons	Pain and fatigue were highly and significantly correlated at 1 and 2 years post-injury (R = 0.49 and 0.62, respectively, P < 0.01)
Cosgrove et al. 1989 [38]	Prospective cohort; 132 TBI patients admitted to inpatient rehab. center	Presence and types of pain	Descriptive	Thirteen patients had peripheral nerve injuries. Four of these 13 patients had pain
Dawson et al. 2007 [58]	Prospective cohort; 94 patients with non-penetrating TBI, 52% with mild TBI, 48% with more severe TBI.	Severity subscale of West Haven–Yale multidimensional pain inventory; avg. 4.3 years post-injury	Multivariate regression	Spearman's rho for association between pain and return to productivity: All TBI: 0.50; Mild TBI:0.45; Mod/Sev. TBI: 0.65. For every unit increase in log of pain, odds of going back to work reduced by 96%. Pain highly correlated with depression in post-hoc analyses.
Guttman 1943 [39]	Retrospective cohort; 158 inpatients with mild to severe TBI	Injury severity, report of headache; admission, during hospital stay, discharge, up to 3—and 6 months post-injury	Descriptive	No apparent difference in headache prevalence by head injury severity at 6 months
Hillier et al. 1997 [40]	Retrospective cohort; 67 mild to severe TBI cases	Self report prevalence of pain/location; 5 years	Descriptive	Fifty-eight percent of pts. had headaches 5 years after TBI. 5% had decreased movement and pain
Hoffman et al. 2007 [33]	Prospective cohort; 146 patients with moderate to severe TBI enrolled in acute inpatient rehab.	Bodily pain (SF-36); 1 year post-injury	Multivariate regression	One year post-injury: 74% of patients had pain, 55% had functional interference. TBI and other injuries were not associated with pain; pain was associated with female gender, lower functioning, non-white, higher depression scores, and lower community involvement. Non-white race and depression remained associated with pain at 1 year in a multivariate regression model.
Masson et al. 1996 [36]	Prospective cohort; adult inpts. with mult. injuries incl. head injury, 231 with TBI. 80 with lower limb injury	Self report of headache and other pain; 5 years post-injury	Bivariate comparisons	Prevalence of headache was 44–54% in head injury patients, and 16% in comparison group (lower limb injury, no TBI). Other pain, but not headache, associated with head injury severity
Nampiaparampil, 2008 [32]	Systematic review of 23 observational studies reporting data on pain syndromes in adult patients with TBI (total n = 4,206).	Pain prevalence	Systematic review of pain prevalence across studies; results stratified by TBI severity	Prevalence of chronic pain was 52% among 3,289 civilian patients with TBI based on 20 studies. Prevalence of pain among 917 veterans with TBI from 3 studies was 43%. Prevalence of headache was 58% among 1,670 patients based on 12 studies. Among civilians, chronic pain was more frequent (73%) in pts. with mild TBI compared with moderate/severe TBI. TBI appears to be independently related to pain, though PTSD may partially mediate that relationship.
Olver et al. 1996 [37]	Retrospective cohort; 103 TBI pts in inpt. rehab. unit	Headaches; 2 years and 5 years post-injury	Bivariate comparisons	Seventy-two percent of participants experienced severe TBI. Headache prevalence increased from 31% at 2 years to 42% at 5 years
Walker et al. 2005 [34]	Prospective cohort; 109 inpatients with moderate to severe TBI	Headache frequency, location, type; degree of incapacitation; and density (frequency × intensity); 6 and 12 months post-injury	Bivariate comparisons	At acute rehab, 38% of pts. reported headache symptoms, 76% with daily frequency. Headache unrelated to injury severity, emotional, or demographic variables. Among pts. with headache at admission, 54% reported persistent headache symptoms at 6 months; of this group, 96% still had headaches at 12 months. Twenty-two percent of sample reported delayed onset of headache at 6 or 12 months. Headache improvement associated with less anxiety and depression at 6 months.

PTSD = Post-traumatic Stress Disorder; SF-36 = 36-Item Short Form Health Survey; VAS = Visual Analog Scale; rehab. = rehabilitation; inpts. = inpatients; mult. = multiple; incl. = including; mod/sev. = moderate to severe.

We also identified a number of studies reporting on the prevalence of pain and headache among TBI patients [33–51]. Headache was the most common pain complaint among TBI patients in one prospective cohort and several cross-sectional studies. In the Nampiaparampil 2008 systematic review described previously, the prevalence of chronic pain was 51.5% in patients with mild TBI, compared with 32.1% among patients with moderate or severe TBI [32]. However, our review identified a number of studies not included in the Nampiaparampil 2008 review, and found mixed results on whether the prevalence of headache varies with severity of head injury. While one fair-quality retrospective cohort study [52], one fair-quality case-control study [42], and five cross-sectional studies [45,48,49,53,54] reported a higher prevalence of headache among patients with mild TBI as compared with more severe TBI, there were no differences in the prevalence of headache according to severity of head injury in one retrospective [39] and four prospective cohort studies [33,34,36,55] of varying quality, as well as three cross-sectional studies [44,56,57]. While a more detailed exploration of this topic is beyond the scope of this manuscript, overall we found only limited evidence showing that patients with mild TBI are more likely to have headache or other pain than patients with more severe TBI.

In one cross-sectional study [43], blast exposure was not found to be a significant predictor of pain. Several cohort studies [33,34,58] and one cross-sectional study with TBI patients [59] identified associations among pain or headache and depression. In another cohort study of TBI patients [35], frequency of chronic pain was associated with PTSD. Finally, one cohort study [60], two case-control studies [41,42], and two cross-sectional studies [61,62] identified associations between pain or headache and fatigue or insomnia. There were mixed findings regarding the association between demographic factors and pain outcomes in TBI patients [33,34,63].

In summary, we found that among patients with TBI, headache is present in one-third to one-half of patients up to 5 years after injury (GRADE: Low). Our review showed mixed findings regarding the association between severity of TBI and pain (GRADE: Very Low). Finally, psychological factors, including depression and PTSD, insomnia and fatigue are associated with pain in TBI patients (GRADE: Low).

Findings—Other injuries in polytrauma patients

Twenty-two cohort, two cross-sectional, and five case-control studies addressed patient factors associated with outcomes in patients with polytrauma other than, or in addition to TBI. Table 2 summarizes the findings from the cohort studies. A number of studies support that injury characteristics (in particular, location, severity, and multiplicity) are often associated with pain-related outcomes in patients with amputations or orthopedic injuries. Thirteen cohort studies of varying quality [9,64–76] and one cross-sectional study [77] found associations between injury severity or multiplicity and subsequent pain and functional outcomes. In contrast, one fair-quality retrospective cohort study [78] and one poor-quality case-control study [79] did not find these associations. Some studies that included patients with cognitive disabilities or head injury found associations between head injury or cognitive disability and decreased functional outcome [73–76,80]. In an additional poor-quality case-control study, patients with head injury plus other types of polytrauma were not found to have worse bodily pain at 6 or 12 months compared with patients with TBI alone [81]. Other injury characteristics associated with worse functional and pain outcomes were lower limb injuries [71,75,82–84], open pelvic fracture as compared with closed pelvic fracture [85], and longer length of hospital stay [76].

Table 2. Key Question 4 Studies: Patients with polytraumatic injuries other than or in addition to traumatic brain injury.

Study	Design and Sample	Pain-Related Outcomes Measured	Analytic Method	Results
Anke et al. 1997 [73]	Retrospective cohort; 69 pts. with severe multiple trauma (Oslo, Norway)	Injury severity, social network, impairments, pain; 35 months post-injury	Descriptive; bivariate comparisons	Twenty-nine percent reported moderate to unbearable pain. Pain was significantly correlated only with loss of non-work activities.
Brenneman et al. 1997 [85]	Retrospective cohort; 27 blunt trauma victims with open or closed pelvic fracture	Bodily pain (SF-36), FIM, and Global health assessment; avg. 4 years post-injury	Bivariate comparisons	Patients scored significantly worse on bodily pain and global health than age-specific population norms. No difference found between open fractures group vs closed fractures group; Bodily pain: 55.6 vs 66.1, P = n.s.
Brenneman et al. 1997 [70]	Prospective cohort; 195 survivors of blunt trauma (Ontario, Canada)	Bodily pain (SF-36); Return to work (RTW); 1 year post-injury	Bivariate comparisons; Logistic regression	Reported bodily pain significantly worse at both discharge and year 1 for those unemployed at year 1.
Dimopoulou et al. 2004 [74]	Prospective cohort; 87 multiple-trauma inpatients (Athens, Greece)	Nottingham Health Profile, Rosser Disability Scale; 1 year post-injury	Descriptive; logistic regression	Forty-one percent of patients had pain 1 year after ICU discharge. Injury severity and severe head trauma predicted both poor health-related quality of life and disability.
Dougherty 1999 [79]	Retrospective cohort; 23 veterans with bilateral above-knee amputations; 145 men who completed national survey	Bodily pain (SF-36); avg. 27.5 years post-surgery	Bivariate comparisons	Amputation group scored lower than control group on physical functioning scale of the SF-36. Comparisons on the other seven subscales (including bodily pain) revealed no significant differences.
Dougherty 2001 [69]	Retrospective cohort; 72 pts who underwent transtibial amputation in Vietnam War	Bodily pain (SF-36); avg. 27.5 years post-surgery	Multivariate comparisons	The patient group with more than one amputation or additional injury, but not the single-amputation group, reported significantly worse pain scores than matched controls.
Fitzharris et al. 2007 [86]	Prospective cohort; 62 inpatients injured in motor vehicle accidents (Australia)	Bodily pain (SF-36) and 100-pt VAS; discharge, 2 and 8 months post-discharge	Multivariate comparisons	Pain scores (VAS) remained significantly higher at 8 months post-accident than pre-accident even though pain improved between 2 months and 8 months post-accident. Bodily pain scores (SF-36) were significantly worse 8 months post accident for males, but not females.
Frink et al. 2007 [78]	Retrospective cohort; 26 patients treated for compartment syndrome	Reported pain during extension exercises; avg. 2.4 years after hospital admission	Bivariate comparisons	No significant differences between polytrauma and single injury in muscular strength, range of motion, or pain.
Hebert & Burnham 2000 [65]	Prospective cohort; 830 inpts with traumatic spine injury with or without neural injury	FIM; 1 and 2 years post-injury	Linear regression	Pts with higher injury severity more likely to have incapacitating pain vs occasional pain at 1 year and more likely to have incapacitating pain vs no or occasional pain at 2 years
Holtslag et al. 2007 [75]	Prospective cohort; 311 severely injured patients, majority polytrauma (Netherlands)	Health status (EQ-5D), disability (GOS), Head injury symptoms, comorbidity; 12 to 28 months after trauma	Multivariate logistic regression	Factors associated with lower EQ-5D scores: Education; Comorbidity; Education; Brain Injury; Spinal cord injury and lower extremity injury. Factors significantly predicting pain scores: comorbidity, lower extremity injury.
MacKenzie et al. 1998 [87]	Prospective cohort; 312 pts. admitted for blunt, unilateral lower extremity fracture distal or including the acetabulum	Physical range of motion and muscle strength (impairment), Time to return to work (RTW), Pain (VAS); prior to discharge, 6 and 12 months post-injury	Cox proportional hazards regression adjusted for effect of pain and impairments.	Pain highly correlated with impairment and RTW. RTW percentages within 12 months of injury were 85%, 73%, and 51%, for VAS mean pain scores of <10, 10–19, and ≥30, respectively. Pain did not a predict RTW after adjusting for impairment.
Meerding et al. 2004 [76]	Prospective cohort; 1,806 patients with injuries including polytrauma (Netherlands)	Health Status (EQ-5D), return to work status and duration of absence; 2, 5, and 9 months after surgery	Multivariate regression	Number of injuries predicted health status and duration of work absence. Pain at 2 months was most prevalent in women, older pts., patients hospitalized ≥7 days, and pts. with hip fracture.
Mkandawire et al. 2002 [71]	Prospective cohort; 158 inpatients, majority polytrauma (Wales)	Functional disability (Bull disability scale), pain; 5 years post-injury	Descriptive	Forty-five percent of patients with single fractures and 91% of patients with multiple fractures had residual disability, while 26% and 60%, respectively, were experiencing moderate or severe pain.
Pezzin et al. 2000 [9]	Retrospective cohort; 78 patients who underwent trauma-related amputation	Injury characteristics and treatment, health and injury problems, bodily pain (SF-36); avg. 7.5 years post-discharge	Descriptive; bivariate comparisons; logistic regression	Twenty-four percent of pts reported severe phantom pain, 24% reported wounds/sores in residual limb in the past month. Seventeen percent reported severe pain in the contralateral limb. Number of nights spent on inpt. rehabilitation service marginally associated with pain score and return to work in adjusted multivariate analyses.
Roganovic and Mandic-Gajic 2006 [67]; Roganovic and Mandic-Gajic 2006 [68]	Retrospective cohort; 326 veterans/active military pts. w. missile-caused nerve injuries (Serbia)	Pain syndromes, pain intensity (before and after tx); post-operative and ≥1 year follow-up	Multivariate regression	Type of pain syndrome, multiple nerve damage, and early onset of pain predict initial pain intensity. Type of pain syndrome, severity of nerve injury, and absence of pain paroxysms predict positive treatment outcomes.
Soberg et al. 2007 [80]	Prospective cohort; 101 patients with multiple injuries, NISS ≥16 (Oslo, Norway)	Bodily pain (SF-36), cognition (COG), disability (WHODAS II), injury characteristics; at injury, discharge, 1 and 2 years post-injury	Bivariate comparisons; multivariate regression	At 2 years, patients reported worse health scores on SF-36 compared with general population; time from injury to return home, social functioning and physical functioning predicted disability.
Stalp et al. 2002 [82]	Prospective cohort; 254 polytrauma patients (Germany)	Pain (MFA, SF-12, HASPOC), disability (GCS), ISS; avg. 2.2 years post-injury	Descriptive; bivariate comparisons	Injuries below knee significantly more limiting; in patients reporting pain and restrictions of the lower extremity: 52% caused by foot and ankle injury, 31% knee or thigh, and 17% femur or hip.
Turchin et al. 1999 [84]	Prospective cohort; 56 matched, multiply injured patients with (group1) and without (group 2) foot injury	Pain (SF-36), Osteoarthritis (WOMAC), foot and ankle trauma (Modified Boston Children's Hospital Grading System); avg. 62 months	Bivariate comparisons	Multiply injured patients with foot injuries reported significantly worse health than patients without foot injuries for each of the three outcome measures.
Ulvik et al. 2008 [52]	Retrospective cohort; 210 polytrauma inpatients (Norway)	Pain/discomfort (EQ-5D); 2–7 years post-injury	Logistic regression	Patients without severe head injury reported more problems with pain/discomfort. Years since trauma was negatively associated with pain/discomfort.
Urquhart et al. 2006 [64]	Prospective cohort; 357 pts with isolated orthopedic injury, 659 with multiple ortho. injuries, 165 w. additional injuries (Australia)	Pain (VAS, SF-12), disability, work status (SIPw); 6 months post-injury	Bivariate comparisons	Marginally significant difference in proportion of patients reporting moderate/severe pain: 47% of additional injuries, 37% of multiple ortho. injuries, and 33% of isolated orthopedic injuries.
Vles et al. 2005 [72]	Prospective cohort; 166 trauma inpatients with an ISS ≥16 (Netherlands)	Pain/discomfort (EQ-5D), disability (GOS), return to work status; 1 year post-injury	Multivariate regression	Factors associated with pain or discomfort (5Q-5D): ISS ≥25; female gender; injury to the chest or thoracic contents; and injury to remaining body areas
Zelle et al. 2005 [66]	Retrospective cohort study; 389 polytrauma pts. with one or more lower-extremity fractures (Germany)	Pain (SF-12), polytrauma outcome (HASPOC), Tegner activity score, inability to work; 10+ years post injury	Multivariate regression	Fractures below the knee joint associated with persisting pain and inability to work.

ANOVA = analysis of variance; COG = cognitive function assessment derived from SF-36; EQ-5D = EuroQol-5D, a measure of health outcome by the EuroQol Group; FIM = Functional Independence Measure; ICU = Intensive Care Unit; SF-12 = 12-item short form Health Survey, a subset of SF-36; NISS = New Injury Severity Score; NHP = Nottingham Health Profile; RTW = Return To Work; SF-36 = 36-Item Short Form Health Survey; VAS = Visual Analog Scale; WHODAS II = World Health Organization; WOMAC = Western Ontario and McMaster Osteoarthritis Index; GOS = Glasgow Outcomes Scale; MFA = Musculoskeletal Function Assessment; HASPOC = Hannover Score for Polytrauma Outcome; SIPw = Work subscale of sickness impact profile.

One good-quality [76] and three fair-quality [9,70,73] cohort studies found associations between younger age and better functional outcomes in polytrauma patients. One good-quality cohort study [76] and one fair-quality cohort study [72] found associations between being male and better functional outcomes. In an additional cross-sectional study, phantom pain was more common in women than men [77]. However, in one fair-quality prospective cohort study, there were no significant gender differences in bodily pain scores on the 36-Item Short Form Health Survey (SF-36) at 2 and 8 months post-injury [86]. Finally, in one fair-quality retrospective cohort study, white race was associated with better SF-36 Physical Component and Bodily Pain outcome scores, higher likelihood of return to work, and lower likelihood of reduced work hours [9]. A number of cohort studies found that higher educational achievement [75,76,87] and having a professional/white collar job [70,73,87] were associated with better functional status. Almost no studies reported on relationships between psychosocial factors and pain-related outcomes in patients with polytraumatic injuries other than TBI.

In summary, characteristics of injuries are associated with clinical outcomes including persistent pain and functional status. Specific factors associated with worse pain-related outcomes include: multiple injuries, foot injuries or injuries below the knee joint, and concurrent head injury or cognitive disability (GRADE: Low). Other factors associated with better outcomes in some studies of patients with polytraumatic injuries other than TBI were younger age, higher educational achievement, having a white collar job, or higher income (GRADE: Very Low).

Key Question 5

What are unique provider and system barriers to detecting and treating pain among polytrauma patients? Have interventions been developed to effectively address these barriers?

Finding

We found no randomized controlled trials, cohort studies, case-control studies, or other systematic observational studies that addressed provider and system barriers to detecting and treating pain among polytrauma patients. One qualitative study of providers from four VA Polytrauma Rehabilitation Centers addressed potential provider and system barriers to treating polytrauma patients [3]. In qualitative interviews, 56 providers reported that polytrauma patients are very complex to treat, and that the work with this population is challenging and emotionally taxing. Increasing use of multidisciplinary and concurrent care, and consultation from experts may be necessary to provide the complex care that is needed.

Discussion

Pain from polytraumatic injuries poses numerous challenges during and after rehabilitation treatment. Pain assessment and intervention efforts are further complicated when the injuries include TBI. The purpose of this project was to identify and synthesize evidence on the assessment and treatment of pain in polytrauma patients. Overall, the literature provides limited evidence to guide clinicians in this area. Although previous reviews indicate that pain may interfere with neurocognitive performance in TBI patients [88–90], we found no studies on how neurocognitive performance may affect measurement of pain or pain-related interference. The literature also provides limited evidence to guide clinicians in selecting among non-surgical pain treatments in patients with polytrauma. Aside from one study indicating that inpatient rehabilitation may improve outcomes among patients with trauma-related amputation [9], and a number of case reports/series [10–31], we did not find any rigorous studies of pain intervention studies in this population. Finally, although several studies have suggested that headache is common among blast injury patients [50,51,91], we found no published studies describing how blast-related headache might differ in terms of phenomenology or treatment from other types of headache pain.

We did find moderate evidence showing that injury factors (including location, severity, and the number of different injuries) are associated with pain and functional status over time. Factors found to be associated with worse outcomes across at least several studies were: multiplicity of injury, head injury or cognitive disability, and lower limb injuries. Factors associated with better outcomes in a few studies were: younger age, higher educational achievement, and having a white collar job. Among TBI patients, factors found to be associated with pain and pain-related function in several studies included depression, PTSD, insomnia, and fatigue.

TBI itself is associated with worse outcomes when compared with polytrauma patients without TBI, and there is some evidence that pain is common among TBI patients, present in one-third to one-half of patients up to 5 years post-injury. Contrary to conventional wisdom and what was documented in a recent systematic review [32], we found very limited evidence to support that patients with mild TBI are more likely to have headache or other pain than patients with more severe TBI. While predominantly cross-sectional studies suggest that patients with mild TBI may be more likely to have headache pain than patients with moderate or severe TBI, six prospective cohort studies and several additional cross-sectional studies did not find a relationship between TBI severity and headache prevalence. We note that most of the cross-sectional studies were done in outpatient settings of patients who had been injured years previously, and did not adjust for potential confounders that may influence relationships between TBI severity and pain. In contrast, most of the patients in the prospective studies were identified in inpatient or rehabilitation settings closer to the time of injury. It is likely that differences in sample composition contribute to the differences in findings between the cross-sectional and cohort studies, such that patients with mild TBI may be more likely to be referred to or attend outpatient follow-up appointments when they have bothersome or persistent symptoms such as headache.

Finally, there is almost no evidence that addresses provider and system barriers to treatment of pain among polytrauma patients. In one rigorously conducted qualitative study, providers reported that polytrauma patients are very complex to treat, and that the work with this population is challenging and emotionally taxing. In order to provide high quality care to this complex patient population, clinicians have increased their use of multidisciplinary and concurrent care, and consultation from experts.

Due to innumerable etiologies and combinations of injuries, there is great heterogeneity among polytrauma patients, and resulting substantial variability among the study samples described in the literature. In addition, samples were drawn from a variety of settings over a wide range of time following injuries. Thus, the conclusions drawn from a particular study or set of studies may have limited relevance for other polytrauma patients or other settings; comparisons must be made cautiously. Because of the degree of heterogeneity among studies and their descriptions, as well as the lack of precision inherent in the term “polytrauma,” we adopted a fairly stringent operational definition for polytrauma and focused on identifying studies that included majorities of patients with polytrauma in their samples. Consequently, one potential limitation of this review is that some manuscripts were not identified that might have relevance for the treatment of pain in polytrauma patients. We note that we did search more broadly for studies pertaining to TBI and blast as there is particular relevance for a large segment of the OEF/OIF patient population. In addition, we did not exclude individual studies based on quality rating alone. Thus, the strength of our conclusions is inherently limited by quality variation among included studies. We did review each cohort and case-control study for overall quality using a rigorously developed approach. Finally, we did not attempt to distinguish studies that reported on patients injured in combat or who were exposed to blast vs studies of patients injured in other settings. It is possible that combat-related trauma is associated with different polytrauma characteristics as well as different pain-related outcomes.

Future Research Recommendations

We asked our technical advisory panel and other experts to review the results of our literature review, suggest potential study topics and designs, and assign priorities to the research topics and designs (Table 3). Priority scores were based on potential to achieve the highest possible impact on patient care in the VA. In rating the suggestions for future research, raters were also asked to consider: 1) the degree to which the proposed research will address information gaps identified in the systematic review; 2) the quantity and quality of the research completed so far including systematic reviews; 3) research currently planned or in progress; 4) the feasibility and timeframe that would be necessary to complete the proposed research; 5) existing barriers that have prevented this research from being undertaken before; and 6) the pros and cons of different research methods that might be appropriate for each research question. Raters gave high priority to a variety of different research designs including randomized trials, prospective observational studies and in some cases, cross-sectional studies. Because these are preliminary rankings, a panel or other mechanism to achieve consensus is needed to refine and finalize the recommendations for future research.

Table 3. Future research topics/designs—ratings of priority.

Key Question or Subquestion	Results of Literature Review	Future Research Topic Suggestions	Median rank 1, 2, or 3 1 = most important	Interquartile range
1. Are pain assessment tools reliable and valid in patients with cognitive deficits due to TBI?	No direct evidence on pain assessment tools in TBI or polytrauma.	Quantitative measurement study of reliability and validity of existing pain assessment tools among patients with varying levels of communicativeness and brain injury.	1	1, 2
		Examine the discriminant validity of tools for distinguishing pain from other forms of distress and impairment (e.g., restlessness, PTSD symptoms)	1	1, 2
		Examine the validity of assessment tools in the context of a clinical trial of an intervention for a specific pain condition. For example, in trial of opioids for patients with multiple orthopedic injuries and TBI, use tool to measure changes in pain over time.	2	1, 2
		Examine utility, reliability and validity of electronic medical record pain assessment modules being used for assessment of non-communicative patients with pain.	2	1, 3
		Qualitative study to identify pain behaviors in different cognitively impaired TBI states. From this information, develop and test new tool or modify existing tools to match severity and type of cognitive impairment.	2	1, 3
		Qualitative/quantitative research using partnership with family (and staff) to identify key pain behaviors. Develop or modify tool to incorporate information from patient, caregivers, and empiric trials of analgesics (i.e., guideline recommended approaches).	2	2, 3
2a. Which treatments improve pain outcomes in polytrauma pts?	Primarily case series and case reports of various pain treatments.	Trials of non-pharmacological interventions of varying treatment intensity, including psychological interventions and telephone-based interventions.	1	1, 2
2b. Which pain treatment approaches enhance overall rehabilitation efforts?		RCTs of integrated treatment approaches including comprehensive interdisciplinary rehabilitation, collaborative care, and treatment involving family members.	1	1, 2
		Compare treatments for common specific core conditions (e.g., TBI; PTSD; Pain) to integrated treatment of these core overlapping symptoms.	1	1, 2
		Systematic prospective observation methods and single case experimental designs with replication to study the relationship between pain control and rehabilitation outcomes.	2	2, 3
3a. Does blast-related headache pain differ from other types of headache pain?	None.	Cross-sectional study describing and comparing characteristics of veterans with headache presumed to be blast-related with veterans with headache not known to be blast-related.	2	1, 2
		Prospective observational cohort study comparing outcomes of veterans with blast-related headache with those without blast-related headache	2	1, 2
		Study associations between comorbid psychiatric conditions and headache among patients with blast-headache; compare with patients with non-blast-headache	2	1, 2
		Perform routine imaging on soldiers exposed to blast to assess for structural abnormalities and correlate with headache symptoms	2	2, 3
3b. Which treatments are best for persistent blast-related headache pain?	None.	Trials of pharmacologic and non-pharmacologic treatments, including patients with blast-related headache in one arm of the trial. Non-pharmacologic interventions to test include cognitive behavioral treatment, hypnosis, relaxation training, and biofeedback.	1	1, 1
4a. What patient factors are associated with pain-related outcomes in polytrauma patients? Have interventions been developed to specifically address these factors? (see 2 above)	Some evidence that injury characteristics, insomnia, fatigue, and psychosocial factors are associated with pain-related outcomes.	Prospective cohort study measuring pain outcomes over time. Measuring contributions of patient characteristics and comorbid conditions, while adjusting for injury characteristics including multiplicity of injury, pain type, and location.	1	1, 2
		Prospective cohort study examining relationships between pain, PTSD, and TBI, and pain-related outcomes.	1	1, 2
		Collaborate with DOD to collect or obtain existing pre-deployment DOD survey data to adjust for baseline characteristics prior to injuries in prospective or retrospective cohort studies.	2	2, 2
		Use data obtained in collaboration with DOD and/or from Landsthuhl Regional Medical Center to identify long-term effects of battlefield, acute-phase, or early rehab. treatment.	2	2, 2
		Evaluate contribution of partnership with family and other social variables to pain-related outcomes in polytrauma patients.	2	2, 3
5a. What are unique provider and system barriers to detecting and treating pain among polytrauma patients? Have interventions been developed to effectively address these barriers?	One qualitative study of interviews with providers.	Establish treatment guidelines for pain in polytrauma based on expert opinion. Disseminate and measure impact of guidelines on care.	1	1, 2
		Evaluate patient perceptions of provider and systems barriers, and the impact of efforts to mitigate those barriers.	2	1, 2
		Evaluate implementation of electronic medical record pain assessment tools in PRCs. Refine tools and reevaluate.	2	1, 2
		Couple evaluation of site-specific organizational factors with multi-site prospective observational study of patient pain-related outcomes. Identify associations between organizational factors and patient outcomes.	2	1, 2
		Identify adaptations VA Polytrauma Rehabilitation Centers are making to accommodate diversity among polytrauma patients with pain.	2	2, 3
		Measure the impact of polytrauma pain education on provider behavior.	2	2, 3
		Formative evaluation of implementation of guidelines and education and impact on treatment processes in PRCs.	2	2, 3

TBI = traumatic brain injury; PTSD = post-traumatic stress disorder; DOD = Department of Defense; PRCs = Polytrauma Rehabilitation Centers.

Conclusion

Overall, there is very little evidence to guide clinicians in assessing pain among patients with cognitive deficits due to TBI, treating pain related to polytrauma, and managing blast-related headache. Some evidence suggests that injury characteristics are associated with pain-related outcomes, and that psychological factors including depression and PTSD, and insomnia and fatigue are associated with pain among TBI patients. Future research is clearly indicated using varying designs to improve our knowledge of pain assessment and treatment in the polytrauma patient population.

Appendix A: Search Strategies

Two librarians independently designed search strategies based on the key questions. The results of both searches were combined into a single reference library.

Search strategy 1

Database: Ovid MEDLINE(R) 〈1950 to January Week 5 2008〉

Search Strategy

1. polytraum$.mp. (2115)

2. exp Multiple Trauma/(7404)

3. (multiple adj3 (wound$ or injur$ or traum$ or casualt$)).mp. (12171)

4. 1 or 2 or 3 (13048)

5. exp Blast Injuries/(1862)

6. exp Brain Injuries/(34104)

7. ((head or crani$ or cereb$ or brain$ or explosi$ or explod$ or blast$) adj3 (traum$ or wound$ or injur$ or damag$)).mp. (88531)

8. 5 or 6 or 7 (90818)

9. exp pain/(218224)

10. exp pain measurement/(33373)

11. exp nociceptors/(8377)

12. (pain$ or agony or agoniz$ or nocicept$).mp. [mp=title, original title, abstract, name of substance word, subject heading word] (334241)

13. 9 or 10 or 11 or 12 (392032)

14. 9 or 10 (230078)

15. 4 and 14 (175)

16. 5 and 14 (12)

17. exp War/(25443)

18. exp Military Personnel/(15657)

19. exp Military Medicine/(21662)

20. exp Veterans/(5122)

21. exp Veterans Disability Claims/(209)

22. Hospitals, Veterans/(4480)

23. exp “United States Department of Veterans Affairs”/(3021)

24. (desert storm or gulf war or enduring freedom or iraqi freedom).mp. (1606)

25. exp Iraq War, 2003 -/or exp Iraq/(2569)

26. (iraq or soldier$ or veteran$ or combat$ or militar$ or battle$).mp. (77729)

27. 17 or 18 or 19 or 20 or 21 or 22 or 23 or 24 or 25 or 26 (94873)

28. exp “wounds and injuries”/or in.fs. (580168)

29. 27 and 28 (9834)

30. 8 and 13 (1988)

31. limit 30 to humans (1836)

32. limit 31 to english language (1402)

33. limit 31 to abstracts (1447)

34. 32 or 33 (1681)

35. 4 and 13 (500)

36. limit 35 to humans (491)

37. limit 36 to english language (359)

38. limit 36 to abstracts (465)

39. 37 or 38 (483)

40. 13 and 29 (449)

41. limit 40 to humans (441)

42. limit 41 to english language (383)

43. limit 41 to abstracts (375)

44. 42 or 43 (424)

45. 34 or 39 or 44 (2480)

46. limit 45 to year = “2000–2008” (1163)

47. limit 45 to year = “1902–1999” (1317)

48. from 47 keep 1 (1)

Search strategy 2

This search was saved in PubMed to provide automatic weekly updates:

• “Brain Injuries”(92)

• OR “Multiple Trauma”(92)

• OR “Blast Injuries”(92)

• OR TBI[All Fields]

• OR “traumatic brain injury”[All Fields]

• OR “traumatic brain injuries”[All Fields]

• OR polytrauma[All Fields]

• OR multitrauma[All Fields]

• OR “multi trauma”[All Fields]

• OR “poly trauma”[All Fields]

• OR ((“Wounds and Injuries”(92) OR “injuries” [Subheading]) AND (“War”(92) OR “Iraq War, 2003 -”(92) OR “Gulf War”(92)))

• AND (“pain”[MeSH Terms] OR pain[Text Word]

Appendix B: USPSTF Quality Rating Criteria

Randomized Controlled Trials (RCTs) and Cohort Studies

Criteria

• Initial assembly of comparable groups: RCTs—adequate randomization, including concealment and whether potential confounders were distributed equally among groups; cohort studies—consideration of potential confounders with either restriction or measurement for adjustment in the analysis; consideration of inception cohorts

• Maintenance of comparable groups (includes attrition, cross-overs, adherence, contamination)

• Important differential loss to follow-up or overall high loss to follow-up

• Measurements: equal, reliable, and valid (includes masking of outcome assessment)

• Clear definition of interventions

• Important outcomes considered

• Analysis: adjustment for potential confounders for cohort studies, or intention-to-treat analysis for RCTs (i.e., analysis in which all participants in a trial are analyzed according to the intervention to which they were allocated, regardless of whether or not they completed the intervention)

Definition of Ratings Based on Above Criteria

Good:	Meets all criteria: Comparable groups are assembled initially and maintained throughout the study (follow-up at least 80 percent); reliable and valid measurement instruments are used and applied equally to the groups; interventions are spelled out clearly; important outcomes are considered; and appropriate attention to confounders in analysis.
Fair:	Studies will be graded “fair” if any or all of the following problems occur, without the important limitations noted in the “poor” category below: Generally comparable groups are assembled initially but some question remains whether some (although not major) differences occurred in follow-up; measurement instruments are acceptable (although not the best) and generally applied equally; some but not all important outcomes are considered; and some but not all potential confounders are accounted for.
Poor:	Studies will be graded “poor” if any of the following major limitations exists: Groups assembled initially are not close to being comparable or maintained throughout the study; unreliable or invalid measurement instruments are used or not applied at all equally among groups (including not masking outcome assessment); and key confounders are given little or no attention.

Appendix C: Criteria for Assigning GRADE for Body of Evidence

High = Further research is very unlikely to change our confidence on the estimate of effect.

Moderate = Further research is likely to have an important impact on our confidence in the estimate of effect and may change the estimate.

Low = Further research is very likely to have an important impact on our confidence in the estimate of effect and is likely to change the estimate.

Very Low = Any estimate of effect is very uncertain.

The GRADE Working Group also suggests using the following scheme for assigning the “grade” or strength of evidence:

Type of evidence

Randomized trial = high

Observational study = low

Any other evidence = very low

Decrease GRADE if

• Serious (−1) or very serious (−2) limitation to study quality

• Important inconsistency (−1)

• Some (−1) or major (−2) uncertainty about directness

• Imprecise or sparse data (−1)

• High probability of reporting bias (−1)

Increase GRADE if

• Strong evidence of association-significant relative risk of >2 (<0.5) based on consistent evidence from two or more observational studies, with no plausible confounders (+1)

• Very strong evidence of association-significant relative risk of >5 (<0.2) based on direct evidence with no major threats to validity (+2)

• Evidence of a dose response gradient (+1)

• All plausible confounders would have reduced the effect (+1)